US Venture Capital Report 2025 — Concentration

The single most useful thing to understand about US venture capital in 2025 is that the aggregate numbers stopped describing the typical experience.

This is not a complaint about data quality. It is a consequence of concentration. When a small number of transactions account for a large share of total value, the total measures those transactions rather than the market. A year in which total deal value rises while the number of companies raising falls is a year in which most companies faced a harder market than the year before — but the headline says the opposite.

That was 2025. Capital concentrated in three dimensions simultaneously:

• By category, into artificial intelligence and adjacent infrastructure.
• By company, into a small number of very large rounds, several of which individually exceeded what entire sectors received.
• By firm, as the largest platforms — with multiple funds, permanent capital, and the ability to lead at any stage — took a rising share of deployment.

Each of these was individually explicable. Together they produced a market where the distinction between the median company's experience and the aggregate statistics became wider than at any point in the industry's modern history.

Beneath this, the structural issue continued to build. Seed formation outside favoured categories was constrained for a fourth consecutive year. At four years, this stops being a cyclical downturn and becomes a structural gap in the population of companies that will exist at growth stage in the late 2020s.

Reading a concentrated market

The measurement problem deserves precise treatment because it affects every claim about the year.

Deal value is the sum of capital deployed. It is dominated by the largest transactions — a single multi-billion-dollar round can exceed the combined total of hundreds of seed rounds.

Deal count is the number of companies that raised. Each company contributes equally regardless of size.

In a market with a normal distribution of round sizes, the two track each other and either works as a summary. In a concentrated market, they diverge, and they diverge in a way that is systematically misleading:

• Value rises because the largest rounds got larger.
• Count falls because fewer companies raised at all.
• The headline reports value, because it is the larger and more quotable number.

The interpretive consequence is direct. For a founder or an early-stage investor, deal count is the relevant statistic. It measures whether companies are getting funded. Deal value measures how much money moved, which in a concentrated market is a statement about a handful of transactions.

When the top ten transactions account for a large share of the total, the total is a statement about ten companies. Everything else in the market is in the residual.

Three practices follow for anyone using 2025 data:

• Report count and value together, always, and treat divergence between them as the signal rather than as noise.
• Use medians rather than means for round sizes and valuations, since means in a concentrated distribution are pulled by the tail.
• Segment before aggregating. A single market-level number for a bifurcated market is a weighted average of two conditions and describes neither.

Firm-level concentration

Less discussed than category concentration, and arguably more consequential for the industry's structure, was the consolidation of deployment into the largest firms.

Several structural advantages compounded through 2025:

• Multi-stage capability. A firm that can lead at seed and continue through growth offers a founder something a single-stage fund cannot: capital continuity without a new process at each stage. In a market where fundraising is costly and uncertain, that has real value.
• Fundraising advantage. With LPs constrained and prioritising established relationships, large firms with long records raised more easily. The constraint that thinned the emerging manager pipeline in 2024 advantaged incumbents by the same mechanism.
• Balance sheet and permanent capital. Some of the largest firms had access to capital structures that did not operate on the traditional ten-year fund cycle, allowing them to hold positions longer and act with less timing pressure.
• Brand as an access asset. In a market where the best companies choose their investors, the ability to win a competitive round is itself a compounding advantage — and it accrues to the firms that already have it.

Each advantage reinforces the others, which is the definition of a compounding structural position rather than a temporary one.

The consequences are worth stating without editorialising, because they cut in more than one direction:

• For founders, the largest firms offer genuine value — continuity of capital, operational resources, signalling.
• For LPs, concentration into large diversified platforms produces returns that increasingly resemble an index of the asset class rather than the concentrated outperformance venture was historically bought for.
• For the industry, a smaller number of larger decision-makers means fewer independent views on what is worth funding — a narrowing of the diversity of bets that venture capital exists to make.

That last point is the substantive concern. Venture capital's economic function is to fund heterogeneous bets on uncertain outcomes. That function depends on many independent decision-makers with different views. Concentration reduces the number of independent views, whatever it does to any individual firm's returns.

The fourth year of the seed constraint

The cohort gap described in the 2025 and 2026 slot-A reports reached a threshold in 2025 that changes its character.

The mechanism, restated: seed formation outside favoured categories has been constrained since 2022. Companies funded at seed reach Series A two to three years later, Series B four to five. A shortfall at seed appears as thin supply at later stages on that lag.

One or two constrained years is a cyclical gap. The affected companies are delayed, some are funded late, and the pipeline broadly recovers.

Four consecutive constrained years is structural, for reasons that go beyond arithmetic:

• The founders who would have started those companies did something else. A person who considered starting a company in 2023 and could not raise took a job. They may start one later, but the specific company and its specific timing are gone.
• The seed funds that would have backed them shrank or disappeared. Fund formation at the small end contracted, and the specialised early-stage capacity that existed in 2021 does not simply reappear when conditions improve.
• The opportunities were addressed by someone else — often an incumbent, often less well, but addressed nonetheless. A market opening does not wait for a company to be funded.

The forward consequence is a specific and predictable distortion, and it is arriving now rather than hypothetically:

A period in which growth-stage capital is available but the population of companies that have reached growth-stage milestones is unusually thin. Too much capital chasing too few qualifying assets raises prices — which looks like a strong market and is actually a supply shortage.

Distinguishing the two requires exactly the discipline the concentration section describes: look at deal count alongside deal value. A thin market with rising prices and a broad market with rising prices look identical in value data and completely different in count data.

Exits: better, not fixed

Listing activity improved from the trough. The improvement was real and it was narrow.

The window admitted companies that were large, profitable or clearly approaching it, in favoured sectors, with sufficient scale to support institutional trading. That describes a small fraction of the backlog accumulated since 2021.

For the rest, the available paths remained what they had been: continue extending runway while growing into the valuation, accept a sale below the last private mark, or restructure. All three occurred, mostly without announcement — which means, as the 2023 report notes, that they are largely absent from the data.

The consequence for the LP liquidity chain is the part that matters. Distributions improved but did not normalise. A partial reopening relieves pressure without resolving it, and for institutional planning purposes that is a materially different situation from either a closed market or a functioning one: it is harder to plan around than either, because it offers enough improvement to defer difficult decisions without providing enough capital to make them unnecessary.

Secondaries and continuation vehicles remained structural, as they had since 2023. Their persistence through a period of partial improvement confirms that they represent a durable change in how private capital manages duration rather than a crisis adaptation.

What firm concentration means for a limited partner

The consolidation of deployment into the largest firms is usually discussed as an industry structure question. It has direct consequences for an LP's expected return, and they cut in more than one direction.

The case for allocating to the largest platforms:

• Access. In a market where the best companies choose their investors, the ability to win a competitive round is itself the scarce asset — and it accrues to the firms that already have it.
• Capital continuity. A firm that can lead from seed through growth offers founders something a single-stage fund cannot, which is a genuine advantage in a market where fundraising is costly and uncertain.
• Resources. Operational support, recruiting, and business development at a scale small funds cannot match.
• Durability. A firm with permanent capital structures and long records is more likely to be there in a decade.

The case against, or at least the caveat:

• Returns converge toward the asset class. A large, diversified, multi-stage platform holds enough companies that its return increasingly resembles an index of venture capital. That is not what venture capital is bought for. An LP paying venture fees for index-like exposure is paying for concentration they are not receiving.
• Fund size changes the required outcome. As the 2018 report establishes, a larger fund needs larger outcomes for the same effect, which narrows the opportunity set and concentrates demand onto a small number of assets.
• Fee economics change the alignment. A firm whose management fee alone supports the partnership has a different relationship to carried interest than one dependent on it.
• The historical record was generated by a different strategy. The returns that justify the allocation frequently came from smaller funds pursuing a narrower approach.

The industry-level concern is separate and is the more substantive one. Venture capital's economic function is to fund heterogeneous bets on uncertain outcomes, and that depends on many independent decision-makers with different views. Concentration reduces the number of independent views, whatever it does to any individual firm's returns — and the emerging manager constraint described in the 2024 report is the mechanism.

An LP can rationally allocate to large platforms for access and durability. What they should not do is expect the return profile that made venture capital attractive, from a vehicle whose size has moved it toward the market's average.

What an allocator could act on

Report count and value together, always. Divergence between them is the signal rather than noise. Value rising while count falls means the largest rounds got larger and fewer companies raised — which for a founder or an early-stage investor is a harder market, reported as a better one.

Use medians, not means. In a concentrated distribution the mean is pulled by the tail and describes nothing.

Segment before aggregating. A single market-level number for a bifurcated market is a weighted average of two conditions and describes neither. The definition of the segment does much of the analytical work, and an approximate segmentation beats an exact aggregate.

Read seed count against a pre-2022 baseline. A recovery relative to a depressed year is not a recovery. The 2019–2021 range is the correct reference for judging whether early-stage formation has repaired.

Distinguish demand strength from supply shortage at Series A. Rising valuations with rising deal count is demand strength; rising valuations with falling count is a thin market. They look identical in valuation data and opposite in count data, and 2026 will produce exactly this ambiguity.

Measure exit improvement by distributions. A partial reopening relieves pressure without resolving it, which is harder to plan around than either a closed or a functioning market — it offers enough improvement to defer difficult decisions without providing enough capital to make them unnecessary.

What 2025 established for US venture

• Concentration made aggregate statistics unreliable, and established deal count as the more informative measure.
• Firm-level consolidation compounded, with structural advantages that reinforce one another.
• The seed constraint became structural at four years, changing the cohort gap from a delay into a permanent shortfall.
• Exit improvement was shown to be narrow, and partial improvement was shown to be harder to plan around than either extreme.
• Secondaries confirmed as permanent by persisting through improving conditions.